Message credits calculation for openai API

Hi community, I have very basic question.
I would like to understand how to count the message credits.
I am using tiktoken python library but that gives count of tokens consumed.
My understanding is as below;
the GPT-3.5 model consumes 1 message credit per interaction, while the GPT-3.5-16k model takes 2 message credits per interaction. The length of the message does not affect the number of message credits consumed.
I got above info from several products offering chatGPT api based solution.

Unable to find the detailed info on how these message credits gets counted and how can I retrieve that info per api key.

Welcome to the forum… pricing is based on input and output tokens. Not sure where you got “1 message credit”… ?


We offer a spectrum of models with different capabilities and price points. In this tutorial, we used text-davinci-003. We recommend using this model or gpt-3.5-turbo while experimenting since they will yield the best results. Once you’ve got things working, you can see if the other models can produce the same results with lower latency and costs. Or if you might need to move to a more powerful model like gpt-4.

The total number of tokens processed in a single request (both prompt and completion) can’t exceed the model’s maximum context length. For most models, this is 4,096 tokens or about 3,000 words. As a rough rule of thumb, 1 token is approximately 4 characters or 0.75 words for English text.

Pricing is pay-as-you-go per 1,000 tokens, with $5 in free credit that can be used during your first 3 months. Learn more.

See also the pricing page…

Or are you asking how other companies created their own credit/token system?

Hi Paul, I understand and therefor I have a confusion. There are several companies are using this term message credit, I’m sharing below Links of the pricing page where they refer to message credits just to give you the context.
Im not sure if this is industry specific term, but I was unable to find related content on openAI docs.


Ah, i was right in my deductions! :wink:

Yeah, you need to come up with your own token/credit/message system. You’ll want to calculate the number of tokens a “message” will take on average and make sure you’re charging more.

1 Like